NIH AI Algorithm Matches Volunteers to Clinical Trials

Press release

Monday, November 18, 2020

This algorithm could save clinicians time and speed up clinical research and recruitment.

The National Institutes of Health has developed a new artificial intelligence algorithm that will help match potential volunteers with relevant clinical trials on the website. ClinicalTrials.gov. The study was published in Nature Communications The AI algorithm called TrialGPT was able to identify clinical trials for which a patient is eligible and then provide a summary explaining how the person fits the study criteria. The researchers concluded that the tool would help clinical staff navigate an ever-changing array of clinical trials for their patients. This could lead to better recruitment for clinical trials and faster progress in medical science.

Read morePrevent upper respiratory infections with ionic air disinfection

Researchers at the National Library of Medicine, NIH, and the National Cancer Institute have leveraged large language models to create an innovative framework that streamlines the clinical trial matching process. TrialGPT processes a patient’s summary, which includes relevant demographic and medical information. It then selects relevant trials from ClinicalTrials.gov for which a person is eligible and excludes trials for which they are not. TrialGPT explains to the patient how they meet the study enrollment criteria. Clinical trials are ranked based on their relevance and eligibility. Clinicians use this list as a tool to discuss clinical trial options with patients.

Stephen Sherry PhD, interim director of the NLM, said: “Machine learning and AI technologies have shown promise for matching patients to clinical trials. However, their practical applications in diverse populations need to be explored further.” This study shows that we can use AI to help physicians more quickly connect patients to a clinical trial that interests them.

The researchers compared TrialGPT's predictions to those made by three clinicians, who evaluated more than 1,000 patient-criterion pairs. The researchers found that TrialGPT was nearly as accurate as three clinicians who evaluated more than 1,000 patient-criterion pairs.

The researchers also conducted a pilot user study in which they asked two clinicians to match six anonymized patient abstracts to six trials. One clinician manually reviewed the abstracts of each patient-trial pair, checked whether they were eligible, and decided whether they were candidates for the clinical trial. Another clinician assessed the eligibility of the same patient-trial pairs using TrialGPT. The researchers found that clinicians using TrialGPT spent 40% less time screening while maintaining the same accuracy.

Clinicians of potential participants are often the first to hear about clinical trials. These opportunities can lead to important medical breakthroughs that will improve your health. Finding the best clinical trials for participants can be a resource- and time-consuming task, slowing down medical research.

Zhiyong Li, PhD, NLM principal investigator and corresponding author, said, “Our study shows that TrialGPT can help clinicians better connect patients to clinical trial opportunities and save time that could be spent on more difficult tasks that require human expertise.”

The research team was selected to participate in the Director's Challenge Innovation Award Further evaluation of the model's performance and fairness in clinical settings. The researchers hope their work will improve clinical trial recruitment and reduce barriers for underrepresented populations to participate in clinical research.

Researchers from the University of Maryland College Park, the University of Illinois at Urbana-Champaign, the University of Pittsburgh and the Albert Einstein College of Medicine in New York collaborated on the study.

The NLM leads the field of research in data science, biomedical informatics, and houses the largest biomedical collection in existence. The NLM supports and conducts research on methods for recording, storing, and retrieving health information, as well as its preservation and communication. The NLM provides resources and tools used by millions of people each year to analyze and access information on molecular biotechnology, environmental health, and toxicology. More information is available athttps://www.nlm.nih.gov.

The National Institutes of Health: The NIH is the national medical research agency of the U.S. Department of Health and Human Services. It consists of 27 institutes and centers. The NIH, the national medical research agency, consists of 27 institutes and centers and is part of the U.S. Department of Health and Human Services. Visit the NIH for more information about its programs and services. www.nih.gov.

NIH…Transforming Discovery into Healthcare^(r)